Deepfake Fraud and Media Provenance: CYFIRMA’s 2026 Forecast Update

Summary

CYFIRMA says deepfake fraud, synthetic-media volume, provenance adoption and regulation are tracking its forecasts, while 2026 fraud losses will test its projection of $5–10B or more annually.

Key points

• CYFIRMA projected annual deepfake fraud losses of $5–10B or more from 2026 onward.
• The article cites $1B in deepfake fraud losses in 2025, up 669% year over year and from $130M over five years.
• It reports deepfake videos rose from 14,000 in 2019 to about 8 million in 2025.
• The article says C2PA had more than 6,000 members, including Google, Meta and OpenAI.
• It says EU AI Act transparency rules for AI-generated content took effect in August 2026 and notes a proposed US Evidence Rule 707 for AI media.
• CYFIRMA recommends content credentials, provenance checks and out-of-band verification; it says year-end 2026 losses will test whether its forecast is on track.